CN106204063A - A kind of paying customer's method for digging and device - Google Patents
A kind of paying customer's method for digging and device Download PDFInfo
- Publication number
- CN106204063A CN106204063A CN201610511933.6A CN201610511933A CN106204063A CN 106204063 A CN106204063 A CN 106204063A CN 201610511933 A CN201610511933 A CN 201610511933A CN 106204063 A CN106204063 A CN 106204063A
- Authority
- CN
- China
- Prior art keywords
- user
- information
- payment
- payment user
- new probability
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/01—Customer relationship services
- G06Q30/015—Providing customer assistance, e.g. assisting a customer within a business location or via helpdesk
- G06Q30/016—After-sales
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Finance (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention discloses a kind of paying customer's method for digging and device, method includes: obtaining sampled data, described sampled data comprises the Back ground Information of user, behavioural information and sequence information；According to described sampled data, training is for determining that user turns the forecast model of new probability, and wherein, described turn of new probability is that user is changed into the probability of paying customer by non-payment user；Back ground Information according to target non-payment user, behavioural information, sequence information and the described forecast model trained, determine that described target non-payment user's turns new probability；Judge that whether described turn of new probability be more than predetermined threshold value；If it is, by described target non-payment user, be defined as the user of paying customer to be changed into.The application embodiment of the present invention, it is possible to excavate the user being likely to become paying customer in non-payment user.
Description
Technical field
The present invention relates to data mining technology field, particularly to a kind of paying customer's method for digging and device.
Background technology
At present, for meeting multi-level demand and the needs of website self-growth of user, the business of a lot of websites is all simultaneously
Covering paid service and non-payment business, its user can be selected to pay or non-payment user according to self-demand.
Non-payment user relatively, paying customer can enjoy more diversified and high-quality service, meet self
Personalized demand.
And for website, paying customer is more conducive to its content producing high-quality, it is provided that high-quality service is with more preferable
Consumer's Experience, thus attract more user, promote the development of himself.That is, for website, paying customer, the most all
Belong to high-quality user, need to carry out the operation of specialty, therefore, how to excavate and non-payment user may be changed into paying customer
User be problem demanding prompt solution.
Summary of the invention
The purpose of the embodiment of the present invention is to provide a kind of paying customer's method for digging and device, to excavate non-payment use
Family is likely to become the user of paying customer.
For reaching above-mentioned purpose, the embodiment of the invention discloses a kind of paying customer's method for digging, described method includes:
Obtaining sampled data, described sampled data comprises the Back ground Information of user, behavioural information and sequence information；
According to described sampled data, training is for determining that user turns the forecast model of new probability, wherein, described turn of new probability
It is changed into the probability of paying customer by non-payment user for user；
Back ground Information according to target non-payment user, behavioural information, sequence information and the described forecast model trained,
Determine that described target non-payment user's turns new probability；
Judge that whether described turn of new probability be more than predetermined threshold value；
If it is, by described target non-payment user, be defined as the user of paying customer to be changed into.
Preferably, described acquisition sampled data, including:
Add up the non-payment user in preset time period and be changed into the user of paying customer by non-payment user, will be united
The Back ground Information of user, behavioural information and the sequence information that meter arrives, is defined as sampled data.
Preferably, described according to described sampled data, training is used for determining that user turns the forecast model of new probability, including:
Described sampled data is divided into positive sample data and negative sample data, and wherein, described positive sample data is by non-
Paying customer is changed into the data that the user of paying customer is corresponding, and described negative sample data are the data that non-payment user is corresponding；
Extract the characteristic information of the described positive sample data user corresponding with negative sample data respectively；
According to the characteristic information of described user, training is for determining that user turns the forecast model of new probability.
Preferably, the described Back ground Information according to target non-payment user, behavioural information, sequence information and the institute trained
State forecast model, determine that described target non-payment user's turns new probability, including:
Back ground Information, behavioural information and sequence information according to target non-payment user, extracts described target non-payment and uses
The characteristic information at family；
Characteristic information according to described target non-payment user and the described forecast model trained, determine that described target is non-
Paying customer turns new probability.
Preferably, described forecast model is iteration decision-tree model.
For reaching above-mentioned purpose, the embodiment of the invention also discloses a kind of paying customer's excavating gear, described device includes:
Obtaining module, be used for obtaining sampled data, described sampled data comprises the Back ground Information of user, behavioural information and orders
Single information；
Training module, for according to described sampled data, training is used for determining that user turns the forecast model of new probability, its
In, described turn of new probability is that user is changed into the probability of paying customer by non-payment user；
First determines module, for Back ground Information, behavioural information, sequence information and training according to target non-payment user
Good described forecast model, determines that described target non-payment user's turns new probability；
Judge module, is used for judging that whether described turn of new probability be more than predetermined threshold value；
Second determines module, in the case of the judged result at described judge module is for being, by non-for described target pair
Expense family, is defined as the user of paying customer to be changed into.
Preferably, described acquisition module, specifically for:
Add up the non-payment user in preset time period and be changed into the user of paying customer by non-payment user, will be united
The Back ground Information of user, behavioural information and the sequence information that meter arrives, is defined as sampled data.
Preferably, described training module, including:
Divide submodule, for described sampled data being divided into positive sample data and negative sample data, wherein, described just
Sample data is to be changed into, by non-payment user, the data that the user of paying customer is corresponding, and described negative sample data are that non-payment is used
The data that family is corresponding；
First extracts submodule, for extracting the feature of the described positive sample data user corresponding with negative sample data respectively
Information；
Training submodule, for the characteristic information according to described user, training is for determining that user turns the prediction of new probability
Model.
Preferably, described first determines module, including:
Second extracts submodule, for Back ground Information, behavioural information and sequence information according to target non-payment user, carries
Take the characteristic information of described target non-payment user；
Determine submodule, for the characteristic information according to described target non-payment user and the described prediction mould trained
Type, determines that described target non-payment user's turns new probability.
Preferably, described forecast model is iteration decision-tree model.
As seen from the above technical solution, embodiments provide a kind of paying customer's method for digging and device, first,
Obtain the sampled data comprising the Back ground Information of user, behavioural information and sequence information；Then, according to described sampled data, instruction
Practicing for determining that user turns the forecast model of new probability, wherein, described turn of new probability is that user is changed into by non-payment user and pays
The probability at expense family；Further according to the Back ground Information of target non-payment user, behavioural information, sequence information and train described pre-
Survey model, determines the new probability that turns of described target non-payment user, and judges whether described turn of new probability is more than predetermined threshold value, as
Fruit is, by described target non-payment user, to be defined as the user of paying customer to be changed into.It can be seen that the application present invention is real
Execute the solution that example provides, it is possible to obtain target non-payment user and be changed into the probability of paying member user, thus excavate
Non-payment user is likely to become the user of paying customer, provides important technical support for subsequent user operation.
Certainly, arbitrary product or the method for implementing the present invention must be not necessarily required to reach all the above excellent simultaneously
Point.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
In having technology to describe, the required accompanying drawing used is briefly described, it should be apparent that, the accompanying drawing in describing below is only this
Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to
Other accompanying drawing is obtained according to these accompanying drawings.
A kind of schematic flow sheet of paying customer's method for digging that Fig. 1 provides for the embodiment of the present invention；
A kind of structural representation of paying customer's excavating gear that Fig. 2 provides for the embodiment of the present invention.
Detailed description of the invention
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Describe, it is clear that described embodiment is only a part of embodiment of the present invention rather than whole embodiments wholely.Based on
Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under not making creative work premise
Embodiment, broadly falls into the scope of protection of the invention.
The embodiment of the invention discloses a kind of paying customer's method for digging and device, be described in detail individually below.
See a kind of schematic flow sheet of paying customer's method for digging that Fig. 1, Fig. 1 provide for the embodiment of the present invention, permissible
Comprise the steps:
S101, it is thus achieved that sampled data, described sampled data comprises the Back ground Information of user, behavioural information and sequence information.
Specifically, it is thus achieved that sampled data, the non-payment user in preset time period can be added up and turned by non-payment user
Become the user of paying customer, the Back ground Information of user, behavioural information and the sequence information that will be counted on, it is defined as hits
According to.
In actual application, it is also possible to according to real needs, non-payment user in self-defined preset time period and by non-pair
Expense family is changed into the user of paying customer, carries out the statistics of user the most again.Such as, for video website, non-payment user
Can refer in preset time period, have viewing behavior but by statistics day never paying customer or payment state
The user of expired one month, non-payment user the user being changed into paying customer can refer to become for the first time in preset time period
For paying customer or this become paying customer distance one month out of date of last time user.
Specifically, the Back ground Information of user can be by user's unsolicited individual Back ground Information, people in the art
Member, it is understood that from the point of view of to a certain extent, the Back ground Information filled in is the most, illustrates that this website is more paid attention to by user, more
Specifically, the Back ground Information of user can include the relevant informations such as sex, age, residence, education degree；The behavior letter of user
Breath can be user's navigation patterns information in preset time period or viewing behavioural information etc., and it has reacted user in multiple dimensions
The situation of website service is enjoyed on degree；The sequence information of user can be user's paying sequence information in preset time period or
Free sequence information, such as participates in the sequence informations such as the movable complimentary paid service obtained, and those skilled in the art can
To be understood by, sequence information can reflect user's degree of dependence to website paid service.
It is assumed that preset time period is on June 10 ,-2016 years on the 10th May in 2016, can be from certain video website back-end data
The data table related in storehouse counts the user 5,000,000 being changed into paying customer within this time period by non-payment user, active
Non-payment user 40,000,000, then, the number comprising Back ground Information, behavioural information and sequence information that this 4500 general-purpose family is corresponding
According to just may determine that as sampled data.
S102, according to described sampled data, training is for determining that user turns the forecast model of new probability, wherein, described turn
New probability is that user is changed into the probability of paying customer by non-payment user.
Specifically, it was predicted that model can be GBDT (Gradient Boosting Decision Tree, iteration decision tree)
Algorithm model；GBDT is the decision Tree algorithms of a kind of iteration, and this algorithm is made up of many decision trees, and the conclusion of all trees adds up
Do final result, its major advantage is that precision is high, need not do the normalized of feature, can automatically carry out feature selection,
Be easily processed missing values, model interpretability is preferable, be suitable for multiple loss function etc..More specifically, forecast model can be
Xgboost (eXtreme Gradient Boosting) algorithm model, it is a kind of C++ realization of GBDT, is different from traditional
GBDT algorithm, loss function has been done the Taylor expansion of second order, and has added regular terms outside object function by Xgboost algorithm
Entirety seeks optimal solution, the complexity of decline and model in order to weigh object function, avoids over-fitting.Additionally, Xgboost
Algorithm is packaged into multiple storehouse, and by directly being adjusted just can using to relevant parameter, this is prior art, the most no longer
Repeat.
In actual application, it will be appreciated by persons skilled in the art that paying customer excavates can be as two classification
Problem processes.Therefore, it can be divided into described sampled data positive sample data and negative sample data, wherein, positive sample number
According to the data corresponding for being changed into the user of paying customer by non-payment user, negative sample data are the number that non-payment user is corresponding
According to；Then, the characteristic information of the positive sample data user corresponding with negative sample data is extracted respectively；Feature further according to user is believed
Breath, training is for determining that user turns the forecast model of new probability.
Specifically, can first carry out the selection of feature, then align sample data and negative sample data are located respectively
Reason, extracts the characteristic information of user, and is converted into the data that can input for training pattern；Then by the number after above-mentioned conversion
It is trained according to corresponding label input Xgboost algorithm model, obtains a preferred forecast model, this forecast model
Output can be set as user and is changed into the probability of paying customer by non-payment user.Wherein, those skilled in the art can manage
Solving, label value corresponding to positive sample data is 1, and label value corresponding to negative sample data is 0.
It is assumed that for certain video website, S101 obtained sampled data, then, spy can be carried out from sampled data
Levy extraction, first carry out the selection of feature, this video website according to own service, from user base information, viewing behavioural information,
Three angles of sequence information propose 182 dimensional features altogether, and wherein user base information proposes 42 dimensional features altogether, can wrap
Include:
Sex (man, female, the unknown, altogether 3-dimensional)；
Age (1 dimension)；
It is registered to become the natural law (1 dimension) of member；
Filling rate (1 dimension), represents the complete situation of user base information solicitation；
Whether fill in cell-phone number (1 dimension)；
Whether fill in mailbox (1 dimension)；
Whether fill in the pet name (1 dimension)；
Residence (line, two wires, three lines, four line cities, other, unknown, totally 6 dimension)；
Education degree (totally 7 dimension such as primary school, junior middle school, senior middle school, junior college, undergraduate course)；
Income situation (less than 1000 yuan, 1001-2000 unit, 2001-3000 unit etc. totally 9 dimension)；
Third party's account access category (totally 11 dimension such as Baidu, Sina, Renren Network, QQ)；
Wherein, viewing behavioural information proposes 135 dimensional features altogether, may include that
Respectively hold VV (Video View, video playback amount) total amount (totally 27 dimension), represent the video playback total amount of each terminal, as
The video playback total amount etc. that mobile terminal, PC (Personal Computer, PC) hold；
Each channel VV total amount (totally 39 dimension), the most each channel represents the program channel of this video website, such as film, TV
Play, variety etc.；
VV on film, paid video and all videos, sky frequency, portion's number, duration (totally 12 dimension)；
The last program request, try, ad click, viewing VIP (Very Important Person, honored guest) video away from
From buy VIP time (point hour and natural law, totally 8 dimension), wherein, VIP refers to that paying customer, VIP video refer to paid video；
Click is skipped advertisement, is liked advertisement number of times (totally 2 dimension)；
The time (natural law, totally 2 dimension) that the last viewing film, any video distance are bought；
Try number of times, video-on-demand times (totally 2 dimension)；
Buy VV amount (totally 4 dimension) weekly of the previous moon；
The VV amount of new film and acute amount (being divided into reach the standard grade a week, two weeks, three weeks, longer four periods, totally 8 dimension)；
PGC (Professionally-generated Content, professional production content) and UGC (User-
Generated Content, user-generated content) VV amount and acute amount (totally 4 dimension)；
The viewing amount of the big film of cinemas film, stock footage and network (3-dimensional altogether), wherein cinemas film, stock footage by
This website determines according to movie show times；
Member account beaching accommodation quantity (totally 1 dimension)；
Having the platform quantity (totally 1 dimension) of viewing behavior, wherein platform refers to that PC, Andriod (Android), TV TV etc. are flat
Platform；
VV amount (totally 22 dimension) on different definitions.
Wherein, sequence information proposes 5 dimensional features altogether, may include that
Free order numbers (1 dimension) in past one month；
Sky card order numbers (1 dimension) in past one month；
Moon card order numbers (1 dimension) in past one month；
Past one month words spoken by an actor from offstage silver card order numbers (1 dimension)；
Program request card order numbers (1 dimension) in past one month.
Then for being obtained each sample data in sampled data by S101, corresponding according to the feature extraction selected
Characteristic information, and be converted into the data that can input.In actual application, it is also possible to sampled data is carried out according to the ratio of 2:1
Dividing, wherein the sample data of 2/3 is as training data, and the sample data of 1/3 is as test data, for verifying the excellent of model
Bad.
It will be appreciated by persons skilled in the art that Xgboost algorithm model can be by data with the form of sparse matrix
Inputting, every data line of matrix all correspond to a user, and concrete pattern of the input can be: member ID
(identification, identity number), feature number: eigenvalue, label；Matrix column number is the number of users added up.
Then using this sparse matrix as input, Xgboost algorithm model is trained, being correlated with according to Xgboost algorithm model
Definition obtains a preferred forecast model, and concrete model training method belongs to prior art, and here is omitted.
In actual application, use Xgboost algorithm model can also to add up and obtain every Wei Te after forecast model trains
Levy the gain that brings and, it will be appreciated by persons skilled in the art that gain that every dimensional feature brings and represent this dimensional feature pair
The contribution of this forecast model number, i.e. prediction user is changed into the importance of paying customer's probability by non-payment user, because of
This, according to the gain of every dimensional feature and, feature importance can be ranked up.Further, it is also possible in each dimension of feature
On carry out user and turn the calculating of new rate, represent that feature is distributed with stacked column graph, the such as longitudinal axis represents and turns a new rate, and transverse axis represents special
Value indicative, by feature visualization, during by checking that this dimensional feature takes different eigenvalues, turns the fluctuations of new rate, just can be clear
Find out to Chu that this dimensional feature is changed into the importance of paying customer's probability for prediction user by non-payment user.Wherein, turn new
Rate by being changed into the ratio shared by the user of paying customer by non-payment user in the user that added up.
It is assumed that the eigenvalue whether feature " fills in telephone number " is 0,1, wherein eigenvalue is that 0 expression user does not fill out
Writing telephone number, eigenvalue is that 1 expression user fill in telephone number, and by being calculated, eigenvalue is to turn new rate 1% when 0,
Eigenvalue is that to turn new rate when 1 be 5%, for the eigenvalue that this dimensional feature is different, turns new rate difference very big, therefore, and this area skill
Art personnel are changed into the general of paying customer it is understood that whether feature " fills in telephone number " for prediction non-payment user
Rate is critically important, i.e. if non-payment user fill in telephone number, then this user is more likely changed into paying customer.
S103, according to the Back ground Information of target non-payment user, behavioural information, sequence information and the described prediction trained
Model, determines that described target non-payment user's turns new probability.
Specifically, according to the Back ground Information of target non-payment user, behavioural information, sequence information and train described pre-
Survey model, determine that described target non-payment user's turns new probability, can be according to the Back ground Information of target non-payment user, behavior
Information and sequence information, extract the characteristic information of described target non-payment user；Then, according to described target non-payment user's
Characteristic information and the described forecast model trained, determine that described target non-payment user's turns new probability.
In actual application, can first obtain the Back ground Information of target non-payment user, behavioural information, sequence information, so
After, the feature selected according to S102, carry out the extraction of characteristic information, and be converted into the data that can input, concrete pattern of the input can
Think: member ID, feature number: eigenvalue.The forecast model trained in data input S102 after converting again, obtains
Target non-payment user is changed into turn new probability of paying customer.
S104, it is judged that whether described turn of new probability be more than predetermined threshold value, if it is, perform S105.
S105, by described target non-payment user, is defined as the user of paying customer to be changed into.
Specifically, a threshold value can be preset, the new probability that turns obtained in S103 is compared with this predetermined threshold value, this
If skilled person is it is understood that this turn of new probability is more than predetermined threshold value, then target non-payment user can be determined
User for paying customer to be changed into.
It is assumed that predetermined threshold value is 50%, S103 the new probability that turns of the non-payment user Xiao Ming obtained is 80%, then,
Xiao Ming can be defined as the user of paying customer to be changed into.
In actual application, it is also possible to multiple non-payment users to be predicted are predicted simultaneously, its characteristic information is turned
Inputted data after change, constitute a sparse matrix, and every data line of matrix all correspond to a user to be predicted, specifically
Pattern of the input can be: member ID, feature number: eigenvalue；Matrix column number is non-payment number of users to be predicted.Then,
By the forecast model that trained in this sparse matrix input S102, obtain each non-payment user to be predicted and be changed into and pay
Turn new probability at expense family.Again for above-mentioned each turn of new probability, it is judged that whether it is more than predetermined threshold value, the most permissible
Non-payment user corresponding for this turn of new probability is defined as the user of paying customer to be changed into.
It is assumed that predetermined threshold value is 50%, and in S103, simultaneously red, little beautiful, little Bai little to non-payment user, little put down into
Go prediction, respectively obtained and turn a new probability 40%, 90%, 18%, 56%, then, then can be defined as treating by little beautiful and little putting down
It is changed into the user of non-payment user.
Application embodiment illustrated in fig. 1, first, it is thus achieved that comprise adopting of the Back ground Information of user, behavioural information and sequence information
Sample data；Then, according to described sampled data, training is for determining that user turns the forecast model of new probability, and wherein, described turn new
Probability is that user is changed into the probability of paying customer by non-payment user；Back ground Information, OK further according to target non-payment user
For information, sequence information and the described forecast model trained, determine that described target non-payment user's turns new probability, and judge
Whether described turn of new probability be more than predetermined threshold value, if it is, by described target non-payment user, be defined as waiting that being changed into paying uses
The user at family.It can be seen that the solution that application embodiment illustrated in fig. 1 provides, it is possible to obtain target non-payment user and change
For the probability of paying member user, thus excavate the user being likely to become paying customer in non-payment user, for subsequent user
Operation provides important technical support.
Corresponding with above-mentioned embodiment of the method, the embodiment of the present invention additionally provides a kind of paying customer's excavating gear.
See a kind of structural representation of paying customer's excavating gear that Fig. 2, Fig. 2 provide for the embodiment of the present invention, with Fig. 1
Shown flow process is corresponding, including:
Obtaining module 201, be used for obtaining sampled data, described sampled data comprises the Back ground Information of user, behavioural information
And sequence information；
Acquisition module 201 shown in the embodiment of the present invention, specifically may be used for:
Add up the non-payment user in preset time period and be changed into the user of paying customer by non-payment user, will be united
The Back ground Information of user, behavioural information and the sequence information that meter arrives, is defined as sampled data.
Training module 202, for according to described sampled data, training is used for determining that user turns the forecast model of new probability,
Wherein, described turn of new probability is that user is changed into the probability of paying customer by non-payment user.
Specifically, the training module 202 shown in the embodiment of the present invention, may include that
Divide submodule, for described sampled data being divided into positive sample data and negative sample data, wherein, described just
Sample data is to be changed into, by non-payment user, the data that the user of paying customer is corresponding, and described negative sample data are that non-payment is used
The data that family is corresponding；
First extracts submodule, for extracting the feature of the described positive sample data user corresponding with negative sample data respectively
Information；
Training submodule, for the characteristic information according to described user, training is for determining that user turns the prediction of new probability
Model.
First determines module 203, for according to the Back ground Information of target non-payment user, behavioural information, sequence information and
The described forecast model trained, determines that described target non-payment user's turns new probability.
Specifically, first shown in the embodiment of the present invention determines module 203, may include that
Second extracts submodule, for Back ground Information, behavioural information and sequence information according to target non-payment user, carries
Take the characteristic information of described target non-payment user；
Determine submodule, for the characteristic information according to described target non-payment user and the described prediction mould trained
Type, determines that described target non-payment user's turns new probability.
Judge module 204, is used for judging that whether described turn of new probability be more than predetermined threshold value.
Second determines module 205, in the case of the judged result at described judge module 204 is for being, by described mesh
Mark non-payment user, is defined as the user of paying customer to be changed into.
Specifically, above-mentioned forecast model can be iteration decision-tree model.
Application embodiment illustrated in fig. 2, first, it is thus achieved that comprise adopting of the Back ground Information of user, behavioural information and sequence information
Sample data；Then, according to described sampled data, training is for determining that user turns the forecast model of new probability, and wherein, described turn new
Probability is that user is changed into the probability of paying customer by non-payment user；Back ground Information, OK further according to target non-payment user
For information, sequence information and the described forecast model trained, determine that described target non-payment user's turns new probability, and judge
Whether described turn of new probability be more than predetermined threshold value, if it is, by described target non-payment user, be defined as waiting that being changed into paying uses
The user at family.It can be seen that the solution that application embodiment illustrated in fig. 2 provides, it is possible to obtain target non-payment user and change
For the probability of paying member user, thus excavate the user being likely to become paying customer in non-payment user, for subsequent user
Operation provides important technical support.
It should be noted that in this article, the relational terms of such as first and second or the like is used merely to a reality
Body or operation separate with another entity or operating space, and deposit between not necessarily requiring or imply these entities or operating
Relation or order in any this reality.And, term " includes ", " comprising " or its any other variant are intended to
Comprising of nonexcludability, so that include that the process of a series of key element, method, article or equipment not only include that those are wanted
Element, but also include other key elements being not expressly set out, or also include for this process, method, article or equipment
Intrinsic key element.In the case of there is no more restriction, statement " including ... " key element limited, it is not excluded that
Including process, method, article or the equipment of described key element there is also other identical element.
Each embodiment in this specification all uses relevant mode to describe, identical similar portion between each embodiment
Dividing and see mutually, what each embodiment stressed is the difference with other embodiments.Real especially for device
For executing example, owing to it is substantially similar to embodiment of the method, so describe is fairly simple, relevant part sees embodiment of the method
Part illustrate.
One of ordinary skill in the art will appreciate that all or part of step realizing in said method embodiment is can
Completing instructing relevant hardware by program, described program can be stored in computer read/write memory medium,
The storage medium obtained designated herein, such as: ROM/RAM, magnetic disc, CD etc..
The foregoing is only presently preferred embodiments of the present invention, be not intended to limit protection scope of the present invention.All
Any modification, equivalent substitution and improvement etc. made within the spirit and principles in the present invention, are all contained in protection scope of the present invention
In.
Claims (10)
1. paying customer's method for digging, it is characterised in that described method includes:
Obtaining sampled data, described sampled data comprises the Back ground Information of user, behavioural information and sequence information；
According to described sampled data, training is for determining that user turns the forecast model of new probability, and wherein, described turn of new probability is for using
Family is changed into the probability of paying customer by non-payment user；
Back ground Information according to target non-payment user, behavioural information, sequence information and the described forecast model trained, determine
Described target non-payment user turns new probability；
Judge that whether described turn of new probability be more than predetermined threshold value；
If it is, by described target non-payment user, be defined as the user of paying customer to be changed into.
Method the most according to claim 1, it is characterised in that described acquisition sampled data, including:
Add up the non-payment user in preset time period and be changed into the user of paying customer by non-payment user, will be counted on
The Back ground Information of user, behavioural information and sequence information, be defined as sampled data.
Method the most according to claim 2, it is characterised in that described according to described sampled data, training is used for determining use
Family turns the forecast model of new probability, including:
Described sampled data is divided into positive sample data and negative sample data, and wherein, described positive sample data is by non-payment
User is changed into the data that the user of paying customer is corresponding, and described negative sample data are the data that non-payment user is corresponding；
Extract the characteristic information of the described positive sample data user corresponding with negative sample data respectively；
According to the characteristic information of described user, training is for determining that user turns the forecast model of new probability.
Method the most according to claim 3, it is characterised in that the described Back ground Information according to target non-payment user, OK
For information, sequence information and the described forecast model trained, determine that described target non-payment user's turns new probability, including:
Back ground Information, behavioural information and sequence information according to target non-payment user, extracts described target non-payment user's
Characteristic information；
Characteristic information according to described target non-payment user and the described forecast model trained, determine described target non-payment
User turns new probability.
5. according to the method described in any one of Claims 1-4, it is characterised in that described forecast model is iteration decision tree mould
Type.
6. paying customer's excavating gear, it is characterised in that described device includes:
Obtaining module, be used for obtaining sampled data, described sampled data comprises the Back ground Information of user, behavioural information and order letter
Breath；
Training module, for according to described sampled data, training is for determining that user turns the forecast model of new probability, wherein, institute
Stating a turn new probability is that user is changed into the probability of paying customer by non-payment user；
First determines module, for according to the Back ground Information of target non-payment user, behavioural information, sequence information with train
Described forecast model, determines that described target non-payment user's turns new probability；
Judge module, is used for judging that whether described turn of new probability be more than predetermined threshold value；
Second determines module, in the case of the judged result at described judge module is for being, described target non-payment is used
Family, is defined as the user of paying customer to be changed into.
Device the most according to claim 6, it is characterised in that described acquisition module, specifically for:
Add up the non-payment user in preset time period and be changed into the user of paying customer by non-payment user, will be counted on
The Back ground Information of user, behavioural information and sequence information, be defined as sampled data.
Device the most according to claim 7, it is characterised in that described training module, including:
Divide submodule, for described sampled data being divided into positive sample data and negative sample data, wherein, described positive sample
Data are to be changed into, by non-payment user, the data that the user of paying customer is corresponding, and described negative sample data are non-payment user couple
The data answered；
First extracts submodule, for extracting the feature letter of the described positive sample data user corresponding with negative sample data respectively
Breath；
Training submodule, for the characteristic information according to described user, training is for determining that user turns the forecast model of new probability.
Device the most according to claim 8, it is characterised in that described first determines module, including:
Second extracts submodule, for Back ground Information, behavioural information and sequence information according to target non-payment user, extracts institute
State the characteristic information of target non-payment user；
Determine submodule, for the characteristic information according to described target non-payment user and the described forecast model trained, really
Fixed described target non-payment user turns new probability.
10. according to the device described in any one of claim 6 to 9, it is characterised in that described forecast model is iteration decision tree mould
Type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610511933.6A CN106204063A (en) | 2016-06-30 | 2016-06-30 | A kind of paying customer's method for digging and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610511933.6A CN106204063A (en) | 2016-06-30 | 2016-06-30 | A kind of paying customer's method for digging and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106204063A true CN106204063A (en) | 2016-12-07 |
Family
ID=57464084
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610511933.6A Pending CN106204063A (en) | 2016-06-30 | 2016-06-30 | A kind of paying customer's method for digging and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106204063A (en) |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107506877A (en) * | 2017-09-30 | 2017-12-22 | 携程计算机技术（上海）有限公司 | OTA platforms are to Forecasting Methodology and system of the shop without room |
| CN107507042A (en) * | 2017-09-15 | 2017-12-22 | 携程计算机技术（上海）有限公司 | Marketing method and system based on user's portrait |
| CN107633326A (en) * | 2017-09-14 | 2018-01-26 | 北京拉勾科技有限公司 | A kind of user delivers the construction method and computing device of wish model |
| CN107844584A (en) * | 2017-11-14 | 2018-03-27 | 北京小度信息科技有限公司 | Usage mining method, apparatus, electronic equipment and computer-readable recording medium |
| CN107862556A (en) * | 2017-12-04 | 2018-03-30 | 北京奇艺世纪科技有限公司 | A kind of put-on method and system of VIP advertisements |
| CN108009926A (en) * | 2017-12-07 | 2018-05-08 | 上海点融信息科技有限责任公司 | Method, information processor and readable storage medium storing program for executing for user's classification |
| CN108228809A (en) * | 2017-12-29 | 2018-06-29 | 广州品唯软件有限公司 | A kind of parameter determination method and device |
| CN108241984A (en) * | 2016-12-23 | 2018-07-03 | 北京国双科技有限公司 | A kind of visitor's sorting technique and device |
| CN108256910A (en) * | 2018-01-10 | 2018-07-06 | 京东方科技集团股份有限公司 | Content recommendation method, device and electronic equipment |
| CN108268555A (en) * | 2017-01-03 | 2018-07-10 | 中国移动通信有限公司研究院 | A kind of information processing method and device |
| CN108681750A (en) * | 2018-05-21 | 2018-10-19 | 阿里巴巴集团控股有限公司 | The feature of GBDT models explains method and apparatus |
| CN108875761A (en) * | 2017-05-11 | 2018-11-23 | 华为技术有限公司 | A kind of method and device for expanding potential user |
| CN108965938A (en) * | 2018-08-03 | 2018-12-07 | 山东大学 | Potential paying customer prediction technique and system in smart television |
| CN109034903A (en) * | 2018-07-27 | 2018-12-18 | 广州视源电子科技股份有限公司 | User's conversion ratio prediction technique and device, computer readable storage medium |
| CN109166012A (en) * | 2018-09-21 | 2019-01-08 | 苏州创旅天下信息技术有限公司 | The method and apparatus of classification and information push for stroke predetermined class user |
| CN109190794A (en) * | 2018-07-27 | 2019-01-11 | 广州视源电子科技股份有限公司 | Behavior sequence cleaning method and device, user's conversion ratio prediction technique and device |
| CN110363575A (en) * | 2019-06-27 | 2019-10-22 | 上海淇毓信息科技有限公司 | A kind of credit user moves branch wish prediction technique, device and equipment |
| CN111191860A (en) * | 2020-04-14 | 2020-05-22 | 北京热云科技有限公司 | Prediction method based on ensemble learning, prediction system and readable storage medium |
| CN112257890A (en) * | 2019-07-22 | 2021-01-22 | 北京易真学思教育科技有限公司 | Data processing method and device and chargeback prediction model training method and device |
| US10990500B2 (en) | 2018-05-18 | 2021-04-27 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for user analysis |
| CN112866760A (en) * | 2021-01-18 | 2021-05-28 | 青岛聚看云科技有限公司 | Content display method, display equipment and server |
| CN113254644A (en) * | 2021-06-07 | 2021-08-13 | 成都数之联科技有限公司 | Model training method, non-complaint work order processing method, system, device and medium |
| CN114862432A (en) * | 2021-02-04 | 2022-08-05 | 武汉斗鱼鱼乐网络科技有限公司 | Target user determination method and device, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102346894A (en) * | 2010-08-03 | 2012-02-08 | 阿里巴巴集团控股有限公司 | Output method, system and server of recommendation information |
| CN105447730A (en) * | 2015-12-25 | 2016-03-30 | 腾讯科技（深圳）有限公司 | Target user orientation method and device |
| CN105469263A (en) * | 2014-09-24 | 2016-04-06 | 阿里巴巴集团控股有限公司 | Commodity recommendation method and device |
| CN105719164A (en) * | 2016-01-21 | 2016-06-29 | 海信集团有限公司 | Paid multimedia resource recommending method and paid multimedia resource recommending device |
-
2016
- 2016-06-30 CN CN201610511933.6A patent/CN106204063A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102346894A (en) * | 2010-08-03 | 2012-02-08 | 阿里巴巴集团控股有限公司 | Output method, system and server of recommendation information |
| CN105469263A (en) * | 2014-09-24 | 2016-04-06 | 阿里巴巴集团控股有限公司 | Commodity recommendation method and device |
| CN105447730A (en) * | 2015-12-25 | 2016-03-30 | 腾讯科技（深圳）有限公司 | Target user orientation method and device |
| CN105719164A (en) * | 2016-01-21 | 2016-06-29 | 海信集团有限公司 | Paid multimedia resource recommending method and paid multimedia resource recommending device |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108241984A (en) * | 2016-12-23 | 2018-07-03 | 北京国双科技有限公司 | A kind of visitor's sorting technique and device |
| CN108268555A (en) * | 2017-01-03 | 2018-07-10 | 中国移动通信有限公司研究院 | A kind of information processing method and device |
| CN108875761B (en) * | 2017-05-11 | 2022-06-28 | 华为技术有限公司 | Method and device for expanding potential users |
| CN108875761A (en) * | 2017-05-11 | 2018-11-23 | 华为技术有限公司 | A kind of method and device for expanding potential user |
| CN107633326A (en) * | 2017-09-14 | 2018-01-26 | 北京拉勾科技有限公司 | A kind of user delivers the construction method and computing device of wish model |
| CN107507042A (en) * | 2017-09-15 | 2017-12-22 | 携程计算机技术（上海）有限公司 | Marketing method and system based on user's portrait |
| CN107506877A (en) * | 2017-09-30 | 2017-12-22 | 携程计算机技术（上海）有限公司 | OTA platforms are to Forecasting Methodology and system of the shop without room |
| CN107844584A (en) * | 2017-11-14 | 2018-03-27 | 北京小度信息科技有限公司 | Usage mining method, apparatus, electronic equipment and computer-readable recording medium |
| CN107862556A (en) * | 2017-12-04 | 2018-03-30 | 北京奇艺世纪科技有限公司 | A kind of put-on method and system of VIP advertisements |
| CN108009926A (en) * | 2017-12-07 | 2018-05-08 | 上海点融信息科技有限责任公司 | Method, information processor and readable storage medium storing program for executing for user's classification |
| CN108228809A (en) * | 2017-12-29 | 2018-06-29 | 广州品唯软件有限公司 | A kind of parameter determination method and device |
| CN108256910A (en) * | 2018-01-10 | 2018-07-06 | 京东方科技集团股份有限公司 | Content recommendation method, device and electronic equipment |
| US10990500B2 (en) | 2018-05-18 | 2021-04-27 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for user analysis |
| CN108681750A (en) * | 2018-05-21 | 2018-10-19 | 阿里巴巴集团控股有限公司 | The feature of GBDT models explains method and apparatus |
| US11205129B2 (en) | 2018-05-21 | 2021-12-21 | Advanced New Technologies Co., Ltd. | GBDT model feature interpretation method and apparatus |
| CN109190794B (en) * | 2018-07-27 | 2021-08-24 | 广州视源电子科技股份有限公司 | Behavior sequence cleaning method and device, and user conversion rate prediction method and device |
| CN109190794A (en) * | 2018-07-27 | 2019-01-11 | 广州视源电子科技股份有限公司 | Behavior sequence cleaning method and device, user's conversion ratio prediction technique and device |
| CN109034903A (en) * | 2018-07-27 | 2018-12-18 | 广州视源电子科技股份有限公司 | User's conversion ratio prediction technique and device, computer readable storage medium |
| CN108965938B (en) * | 2018-08-03 | 2020-03-20 | 山东大学 | Method and system for predicting potential pay users in smart television |
| CN108965938A (en) * | 2018-08-03 | 2018-12-07 | 山东大学 | Potential paying customer prediction technique and system in smart television |
| CN109166012B (en) * | 2018-09-21 | 2021-05-28 | 苏州创旅天下信息技术有限公司 | Method and device for classifying users in travel reservation class and pushing information |
| CN109166012A (en) * | 2018-09-21 | 2019-01-08 | 苏州创旅天下信息技术有限公司 | The method and apparatus of classification and information push for stroke predetermined class user |
| CN110363575A (en) * | 2019-06-27 | 2019-10-22 | 上海淇毓信息科技有限公司 | A kind of credit user moves branch wish prediction technique, device and equipment |
| CN112257890A (en) * | 2019-07-22 | 2021-01-22 | 北京易真学思教育科技有限公司 | Data processing method and device and chargeback prediction model training method and device |
| CN111191860B (en) * | 2020-04-14 | 2020-08-21 | 北京热云科技有限公司 | Prediction method based on ensemble learning, prediction system and readable storage medium |
| CN111191860A (en) * | 2020-04-14 | 2020-05-22 | 北京热云科技有限公司 | Prediction method based on ensemble learning, prediction system and readable storage medium |
| CN112866760A (en) * | 2021-01-18 | 2021-05-28 | 青岛聚看云科技有限公司 | Content display method, display equipment and server |
| CN114862432A (en) * | 2021-02-04 | 2022-08-05 | 武汉斗鱼鱼乐网络科技有限公司 | Target user determination method and device, electronic equipment and storage medium |
| CN113254644A (en) * | 2021-06-07 | 2021-08-13 | 成都数之联科技有限公司 | Model training method, non-complaint work order processing method, system, device and medium |
| CN113254644B (en) * | 2021-06-07 | 2021-09-17 | 成都数之联科技有限公司 | Model training method, non-complaint work order processing method, system, device and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106204063A (en) | A kind of paying customer's method for digging and device | |
| CN108364085B (en) | Takeout delivery time prediction method and device | |
| Hoque et al. | Adoption of information and communication technology for development: A case study of small and medium enterprises in Bangladesh | |
| Aizaki | Basic functions for supporting an implementation of choice experiments in R | |
| CN106294830A (en) | The recommendation method and device of multimedia resource | |
| CN102811371B (en) | The method, system and device that intelligent television application program is recommended | |
| CN101563701A (en) | Information providing system | |
| CN106339507B (en) | Streaming Media information push method and device | |
| US20150278350A1 (en) | Recommendation System With Dual Collaborative Filter Usage Matrix | |
| CN108681970A (en) | Finance product method for pushing, system and computer storage media based on big data | |
| CN103971257B (en) | The method of estimation and system of effect are launched in a kind of the Internet media combination | |
| Choi et al. | Strategic demand forecasts for the tablet PC market using the Bayesian mixed logit model and market share simulations | |
| CN104102648A (en) | User behavior data based interest recommending method and device | |
| CN104156390A (en) | Comment recommendation method and system | |
| CN106168980A (en) | Multimedia resource recommends sort method and device | |
| CN105488366A (en) | Data permission control method and system | |
| JP6819355B2 (en) | Recommendation generation | |
| CN109543132A (en) | Content recommendation method, device, electronic equipment and storage medium | |
| CN109461030A (en) | A kind of O2O evaluation method and device based on block chain | |
| CN108475256A (en) | Feature insertion is generated from homologous factors | |
| CN108259949A (en) | Method, apparatus and electronic equipment are recommended in a kind of advertisement | |
| CN107205042A (en) | A kind of accurate information method for pushing and system | |
| CN109598404A (en) | Automatically to the method and apparatus for issuing the progress data processing of sales task list | |
| CN109615504A (en) | Products Show method, apparatus, electronic equipment and computer readable storage medium | |
| AU2010212344B2 (en) | Object customization and management system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161207 |
|
| RJ01 | Rejection of invention patent application after publication |